Schedule a Call
Schedule a Call

Reimagining Talent Assessment for an
AI-Enabled World.

AI is opening new possibilities for how we assess, develop and understand talent.

At Think Talent, we are building AI into our assessments to make them more realistic and more scalable, while keeping assessment quality, human judgement and responsible use at the centre.

Explore our AI solutions Our approach to AI

AI can change how we assess. It doesn't change what good assessment requires.

  • Enrich the experience

    Participants step into a live conversation with a character that reacts to what they actually say, instead of choosing from options on a screen.

  • Expand the possibilities

    The same conversation for every participant, at any hour. Hundreds of people can be observed as consistently as ten, and patterns across a whole cohort become visible.

  • Keep assessment quality at the centre

    Every exercise is designed against your competency framework, and trained assessors evaluate every transcript. AI runs the conversation. People make the judgement.

Bringing AI into real assessment situations.

Realistic workplace conversations, with a character that reacts.

Participants handle a challenging conversation with an AI character that holds its brief and its mood. It warms when its concern is heard and hardens when it is brushed past.

The problem
Roleplays show how people really handle difficult moments, but each one needs a trained roleplayer, a time slot and a consistent performance. That limits how many people can take part, and how often.
Where it's used
Preparing first-time managers for their first difficult conversations, practice between the sessions of a learning journey, and live exercises in development centres.
What you gain
A full transcript of every conversation, with turn-level notes, for your assessors to evaluate against your framework.

Runs by chat, in English.

Talk to us about Dialogue
DialogueTIME LEFT 15:00

Brief

  • Meridian Corp. A senior analyst's year-end rating has come in lower than expected, and nobody raised a concern all year.
  • You play: the analyst's manager.
  • The character is: disappointed, anxious about the relationship, and not about to accept a vague answer.

Fixed script, for illustration. Nothing here is recorded.

  1. Analyst · AI characterI've read the rating twice. In twelve months nobody told me anything was off, so I'm trying to understand what changed.
  2. ManagerI know this is disappointing. The rating reflects the delays on the client migration.Assessor note · explains the rating; no question asked yet
  3. Analyst · AI characterThe migration slipped because the scope changed twice. I flagged that in March. If it was a problem, why am I only hearing about it now?
  4. ManagerThat's fair. I should have raised it with you when it happened. What would help you most over the next six months?Assessor note · acknowledges own part; asks an open question
  5. Analyst · AI characterThank you for saying that. Regular check-ins, and knowing early if something isn't landing.

Structured interviews, delivered the same way for everyone.

An AI interviewer asks your questions and probes until each answer is complete: a behavioural event interview, a competency-based interview or your own format. Participants speak naturally and review the transcription before it is sent.

The problem
A structured interview is only as consistent as the interviewer on the day. Probing varies from one interviewer to the next, and so does the evidence collected.
Where it's used
Development and assessment centres, promotion assessments, and large cohorts where interviewer time is the bottleneck.
What you gain
A structured transcript for every participant, ready for trained assessors to evaluate. The AI never scores anyone.

Illustration. Names are invented.

Talk to us about AI Interaction
AI InteractionTIME LEFT 20:00
AI

AI character

Listening…Thinking…Speaking…

You · Rohan Mehta
Mic onNotesTranscriptEnd & submit

Every scenario and interview template is built by Think Talent's assessment team around your roles and competencies, down to the character's name.

What becomes possible when we can observe behaviour at scale?

Every manager met the same difficult moment, and every word they said is in the transcript. Not a sample seen by a panel: the whole group, in its own words.

Who
112 first-line managers at an IT services company, invited to an optional Dialogue conversation.
The scenario
A team member disputes an "average" rating after a year without critical feedback. The failure is the manager's own.
The conditions
Unsupervised, at any hour, over 52 days.

Engagement

96%

of 112 managers held a real conversation (three or more exchanges), with no one supervising. The median ran 23 minutes and 458 words, and 14% started between midnight and 6am.

Behaviour

54%

did not acknowledge a failure of their own; 46% did. A panel sees a handful of managers a day. Here, ownership was observed across the whole cohort, in each manager's own words.

  • Questions dried up.

    Managers opened by asking: one message in four was a question. By the close, almost none were. Listening gave way to solving.

  • When the mood turned, it turned early.

    In 53 of 112 conversations the character warmed to the manager, usually by the seventh exchange. In 15 it turned cold, usually by the fourth: a conversation that fails does so before the manager has finished making their case. The rest held steady.

  • Never once out of character.

    Across 1,353 messages the character held its role and its brief every time, replying in a median of 7 seconds.

Where validation stands. These markers come from language analysis of every transcript, read the same way for every manager. The next stage checks that reading against trained assessor ratings, then against on-the-job performance, and we publish each result as it lands.

Innovation needs to be matched by rigour.

Using AI in assessment takes more than getting the technology to work. These are the questions we hold our own work to, and how we answer each one today.

  1. Validity

    Are we measuring what we intend to measure?

    Your framework defines what good looks like. The AI gathers the evidence.

    Every scenario and interview is designed by our assessment team against your competency framework, with the same discipline as our development-centre exercises. Next in validation: agreement between AI-derived measures and trained assessor ratings, published as it lands.

  2. Reliability

    Can we generate dependable, comparable evidence?

    Same character, same brief, same questions, for every participant.

    Rules decide the character's next move, and its lines come from a library written for it. Across 1,353 messages in 112 unsupervised conversations, it never broke role, replying in a median of 7 seconds.

  3. Human judgement

    Where should AI support the process, and where should people stay responsible?

    The AI never scores anyone.

    AI runs the conversation and produces the transcript. Trained assessors evaluate it and make every judgement about a participant.

  4. Fairness and trust

    How do we address bias, accessibility and data protection?

    No language model ever sees who a participant is.

    Names, emails and employee IDs are replaced with pseudonyms before any model sees a message. Data is encrypted at rest and in transit, never used to train anyone else's model, and kept on the terms of your agreement, in line with India's DPDP Act, 2023. Participants can speak or type, and review each answer before it is sent.

    Read our privacy notice

What AI does here, and what it doesn't.

AI does

  • Runs the conversation or interview.
  • Transcribes speech, which the participant reviews before sending.
  • Masks identifiers before any model sees them.
  • Phrases the character's replies from a library written for it.
  • Surfaces patterns across a cohort's transcripts, for people to interpret.

People do

  • Design every scenario, character and interview.
  • Evaluate every transcript against your framework.
  • Make every decision about a participant.

Exploring how AI can strengthen your assessment process?

Whether you are looking to scale assessments, create richer simulations or understand the implications of AI for your talent strategy, let's explore what is possible.

Talk to our team

info@thinktalentindia.com

+91-9910955257

+91-8828158509